For therapy with potentially toxic drugs, such as those used in infectious diseases, transplantation, and cancer, the medical community routinely accepts enormous interpatient variability in drug exposure that would be totally unacceptable in any other scientific or industrial discipline. One usually gives one size fits all standard therapy and monitors for lack of clinical effect or toxicity. Instead, a clinician can plan in advance to maximize effect and minimize toxicity by setting individualized clinical targets of serum drug concentration or effect for each patient, and then calculating doses to hit the desired targets with the greatest precision. For example, maximum aposteriori probability (MAP) Bayesian approaches have achieved at least partial control of interpatient drug exposure. They have improved quality of care, reduced complications, shortened hospital stay, and lowered costs. However, MAP Bayesian methods are unfamiliar to most clinicians, and they manage only the single most likely version of the patient. They must assume that the proposed dosage regimen will hit the target exactly. They have no way to evaluate, control, and minimize the error with which the target can be hit in the real world. Our project has four aims. 1) We will truly optimize patient drug exposure by combining, for the first time, control of the dose and the timing of the serum concentration measurements (active dual control) in our multiple model (MM) Bayesian software designed for use by clinicians. 2) To more accurately reflect reality, we will further update the software to estimate process noise in the therapeutic environment (e.g. errors in dose timing) separately from measurement noise. We will evaluate this as an objective index of quality for pharmacokinetic studies. 3) To directly test the clinical value of our methods, we will prospectively compare the percentage of patients having therapeutic vancomycin serum concentrations after current standard dosing, after dosing with our software in both its current state and after it is updated with the capabilities proposed here. Secondary endpoints will include therapeutic outcomes and costs. 4) Finally, we propose to extend our goal-oriented MM to explore the dose-response relationship of a new drug after it is given to humans for the first time in phase I/II studies, with the twin goals of defining effective, safe doses in the shortest time and fewest patients.